4

In Azure, I have two Ubuntu VMs running the same python bottle services under one Cloud Service. The two VMs are set to be in the same Load-Balanced Set and are load-balanced in a Round-Robin fashion under that cloud service.

However, my problem is that each request takes 3~8 seconds to get to the underlying VM. (i.e. The load balancer is working, but it's very slow routing the requests.) I assume it's not normal, is it?

It seems that I can't even access the cloud service instance(load balancer) in Azure. Can anyone give me a hint where I might have set wrong and resulted in this?

Thanks!

Md Farid Uddin Kiron
  • 16,817
  • 3
  • 17
  • 43
Jerry Chou
  • 41
  • 3
  • How do you know the requests take 3-8 seconds? What is your methodology for timing them? – Nathan Dec 22 '13 at 20:56
  • Hmm it's a rough estimate, I triggered a request and monitor logs of VMs, which output something right after receiving requests. The duration was generally 3~8 seconds. If I remove one VM from the cloud service(only one left), I can see response log pop out immediately right after I triggered request. So I guess the problem is in the load balancer. – Jerry Chou Dec 23 '13 at 03:30
  • 1
    I'm seeing the same thing with a basic TCP server running on two VMs in Azure and the client on my laptop, load balanced it takes 2+ seconds, as slow as 15 seconds, but direct it takes 100-200ms. I'm running Windows 2012 R2 VMs, but I agree, the issue seems to be related to the Load Balancer. – markus101 May 27 '14 at 18:15
  • You would be unable to diagnose anything here I don't think as the load balancer is a service. I would recommend logging a ticket through the portal for it to be investigated. – Martyn C Nov 16 '16 at 16:08

0 Answers0